inference accuracy of deep neural networks (DNNs) is a crucial performance
metric, but can vary greatly in practice subject to actual test datasets and is
typically unknown due to the lack of ground truth labels.
本文提出了一种基于 Correctness Ranking Loss 的深度神经网络训练方法,可以对类别概率进行显式的正序排名,提高置信度预测。该方法易于实现,不需要额外的计算代价,适用于现有体系结构,并且在分类基准数据集上表现良好。同时,还对置信度估计相关的任务,如超出分布检测和主动学习,具有相同的可靠性。